Accessibility of Electric Vehicles in Washington State

by Alyssa Adair and Sophia Smolinski

Summary

1. How accessible is charging infrastructure for electric vehicles across different regions in Washington State?
Public charging stations and infrastructure are very accessible to electric vehicle owners in urban areas, while more rural areas like Eastern Washington have a lack of publicly available charging stations.
2. Are there areas with a higher concentration of electric vehicles but limited charging infrastructure?
Most definitely. Almost the entire eastern side of Washington state along with areas near the Oregon border and near the Washington coast seem to have a lower concentration of charging stations.
3. Is there a racial disparity in access to electric vehicles in Washington state?
There seems to be. In areas with higher white populations, there is higher EV ownership. In areas with higher populations of POC, EV ownership is lower.
4. Can we identify clusters or hotspots of electric vehicle adoption in specific urban or rural areas?
There seems to be a pattern of electric vehicle adoption in urban areas where there is a higher concentration of people living together. However, viewing the map of electric vehicles, you can see that the rest of Washington has electric vehicles dispersed throughout the state in differing regions.

Motivation

As informed by Fuel Economy (.gov), gas-powered vehicles release about 1.5 billion tons of greenhouse gas emissions into the atmosphere each year. With the rapid manufacturing, production, and selling of electric vehicles, along with growing government policies promoting and potentially transitioning our society to the use of electric vehicles over gas-powered vehicles, we find this topic to be very relevant. Many hail electric vehicles as one of the most important ways individuals as consumers can contribute to reducing greenhouse gas emissions, yet access to these electric vehicles is often a socioeconomic privilege. We are curious as to how race can affect access to electric vehicles. Additionally, we are interested in how accessible charging stations are and if there are areas where there is a lack of charging infrastructure.

Data Setting

Electric Vehicle Population Dataset URL: https://catalog.data.gov/dataset/electric-vehicle-population-data

This dataset shows the Battery Electric Vehicles (BEVs) and Plug-in Hybrid Electric Vehicles (PHEVs) that are currently registered through the Washington State Department of Licensing (DoL). This dataset is continually updated, so the data for EVs in Washington is present-day.

EV Charging Station Location Dataset URL: https://afdc.energy.gov/fuels/electricity_locations.html#/find/nearest?location=WA

This is one of two extra datasets we are including to meet a challenge goal. This dataset consists of EV charging stations and their corresponding location (coordinates, city, zip code) along with some other information on the chargers that we may not use, depending on insights and finding answers to our research questions. Some unimportant columns for using this dataset in our context would be columns like payment type, station phone number, and owners, to name a few.

2010 Census Race Data by Legislative District URL: https://data.wa.gov/Demographics/WAOFM-Legislative-Districts-Table-2-Census-2010-Po/di4y-k8za/about_data

By including this extra dataset, we are able to meet one of our challenge goals. It includes racial demographic data from the 2010 census in Washington, organized by Legislative District. The reason this dataset is being used is that the electric vehicle data also includes what legislative district the vehicle is registered in, so this will be a useful column to merge on later for comparison. Since this is census data that only focuses on a specific year, it may be hard to draw many meaningful conclusions, especially when the times when data was collected do not match between the EV data and this dataset, but I still think it will be interesting to compare legislative district racial demographics with EV ownership in the area.

Method

2010 Census Race + EV dataset methods

• Load in the racial demographics and EV datasets.
• Clean the datasets by dropping NA rows, and making sure that column types are the same for merging.
• Calculate the percentage EV ownership and racial distributions.
• Merge the datasets into a data frame by the column Legislative District.
• Use plotly express to create an interactive scatterplot showing the relationship between the percentage of white individuals to the percentage of EV ownership per legislative district.
• Use seaborn to create a barplot representing the racial demographics of the legislative districts with the highest and lowest EV ownership.

Charging station + EV dataset methods

• Load in the charging stations dataset and the EV dataset.
• For the charging station dataset, the columns latitude and longitude are provided, which signal a location. However, this is not a format that geopandas can identify. Therefore, we need to create a geometry column, using geopandas to combine latitude and longitude.
• Develop map visualizations for EV locations relative to charging stations using geopandas.
• Also, develop map visualization for EV locations based on Legislative District relative to charging stations.
• Use a new library such as folium or plotly to achieve our second challenge goal, to make the map interactive. One example of how to make this map interactive would be the ability to hold your cursor over a specific point, and it would give data on the location of either an electric vehicle or a charging station.

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
!pip install plotly
import plotly.express as px
import numpy as np
import geopandas as gpd
from shapely import wkt
from shapely.geometry import Point
!pip install -q folium mapclassify
!pip install -q plotly
import folium
from folium.plugins import MarkerCluster
from IPython.display import display
from IPython.display import HTML

Requirement already satisfied: plotly in /opt/conda/lib/python3.10/site-packages (5.19.0)
Requirement already satisfied: tenacity>=6.2.0 in /opt/conda/lib/python3.10/site-packages (from plotly) (8.2.3)
Requirement already satisfied: packaging in /opt/conda/lib/python3.10/site-packages (from plotly) (23.2)

"""Imports the WA Electric Vehicle and 2010 Census Race Data by Legislative District datasets"""
# Read in EV dataset
ev = pd.read_csv('ev.csv')
# Read in census data
leg = pd.read_csv('legislative.csv')

dirty_charging_df = pd.read_csv("EV_Charging_Stations.csv")

clean_charging_df = dirty_charging_df[['Fuel Type Code', 'Station Name', 'Street Address',
       'City', 'State', 'ZIP', 'Station Phone', 'Groups With Access Code',
       'Access Days Time', 'EV Network', 'Geocode Status',
       'Latitude', 'Longitude', 'Date Last Confirmed',
       'ID', 'Updated At', 'EV Connector Types', 'Country',
       'Access Code', 'EV Workplace Charging']]

ev_df = pd.read_csv("Clean_EV_Population.csv")
charging_df = pd.read_csv("Clean_EV_Charging_Stations.csv")
racial_df = pd.read_csv("Clean_Legislative.csv")
state_data = gpd.read_file("tl_2010_53_tract00.shp")

charging_df['geometry'] = [Point(xy) for xy in zip(charging_df.Longitude, charging_df.Latitude)]

racial_df['Legislative District'] = pd.to_numeric(racial_df['Legislative District'], errors='coerce')

"""Cleans the EV dataset"""
# Dropping NA values in the Vehicle Location and Legislative District columns
ev.dropna(subset=['Vehicle Location', 'Legislative District'], inplace=True)
# Legislative District column in EV dataset was a float type. This converts the column to a string type
ev['Legislative District'] = ev['Legislative District'].astype(int)
ev['Legislative District'] = ev['Legislative District'].astype(str)

"""Percent EV ownership per legislative district"""
# Counts the number of electric vehicles in each legislative district, then calculates the sum of all the EVs
# in the data set
evs_per_district = ev.groupby('Legislative District').size()
total_evs = evs_per_district.sum()
# Calulcating the percentage of EVs per district out of all EVs in the state, rounding to 2 decimal points
rates = (evs_per_district / total_evs) * 100
rates = rates.round(2)
# Resets the index to aid in merging the two data frames
rates = rates.reset_index()
rates.columns = ['Legislative District', 'Percentage EV Ownership']

# Having an issue with the merge function creating duplicate 'Percentage EV Ownership' columns when
# the cell was run multiple times. This drops the column in the EV dataset to avoid this issue.
if 'Percentage EV Ownership' in ev:
    ev.drop(columns='Percentage EV Ownership', inplace=True)

ev = pd.merge(ev, rates, on='Legislative District', how='left')

# Testing the Percentage EV Ownership calculations on a small subset of the dataframe
rates_subset = rates[rates['Legislative District'].isin(['5', '6', '7', '8', '9'])]
rates_subset = rates_subset.reset_index(drop=True)
expected_output1 = pd.DataFrame({'Legislative District': ['5', '6', '7', '8', '9'], 'Percentage EV Ownership': [4.24, 0.96, 0.47, 1.04, 0.56]})
assert rates_subset.equals(expected_output1), 'Test failed'
print('Test passed')
ev

Test passed

	VIN (1-10)	County	City	State	Postal Code	Model Year	Make	Model	Electric Vehicle Type	Clean Alternative Fuel Vehicle (CAFV) Eligibility	Electric Range	Base MSRP	Legislative District	DOL Vehicle ID	Vehicle Location	Electric Utility	2020 Census Tract	Percentage EV Ownership
0	3C3CFFGE4E	Yakima	Yakima	WA	98902.0	2014	FIAT	500	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	87	0	14	1593721	POINT (-120.524012 46.5973939)	PACIFICORP	5.307700e+10	0.65
1	5YJXCBE40H	Thurston	Olympia	WA	98513.0	2017	TESLA	MODEL X	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	200	0	2	257167501	POINT (-122.817545 46.98876)	PUGET SOUND ENERGY INC	5.306701e+10	1.13
2	3MW39FS03P	King	Renton	WA	98058.0	2023	BMW	330E	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	20	0	11	224071816	POINT (-122.1298876 47.4451257)	PUGET SOUND ENERGY INC||CITY OF TACOMA - (WA)	5.303303e+10	3.97
3	7PDSGABA8P	Snohomish	Bothell	WA	98012.0	2023	RIVIAN	R1S	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	0	0	21	260084653	POINT (-122.1873 47.820245)	PUGET SOUND ENERGY INC	5.306105e+10	2.50
4	5YJ3E1EB8L	King	Kent	WA	98031.0	2020	TESLA	MODEL 3	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	322	0	33	253771913	POINT (-122.2012521 47.3931814)	PUGET SOUND ENERGY INC||CITY OF TACOMA - (WA)	5.303303e+10	2.03
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
166430	3FA6P0SU4D	Spokane	Spokane	WA	99223.0	2013	FORD	FUSION	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	19	0	6	239527123	POINT (-117.369705 47.62637)	BONNEVILLE POWER ADMINISTRATION||AVISTA CORP||...	5.306300e+10	0.96
166431	5YJYGDEE5M	King	Sammamish	WA	98074.0	2021	TESLA	MODEL Y	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	0	0	45	148715479	POINT (-122.0313266 47.6285782)	PUGET SOUND ENERGY INC||CITY OF TACOMA - (WA)	5.303303e+10	6.05
166432	7SAYGDEE5N	Snohomish	Mukilteo	WA	98275.0	2022	TESLA	MODEL Y	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	0	0	21	220504406	POINT (-122.299965 47.94171)	PUGET SOUND ENERGY INC	5.306104e+10	2.50
166433	1G1RH6E43D	Lewis	Mossyrock	WA	98564.0	2013	CHEVROLET	VOLT	Plug-in Hybrid Electric Vehicle (PHEV)	Clean Alternative Fuel Vehicle Eligible	38	0	20	156418475	POINT (-122.487535 46.5290135)	BONNEVILLE POWER ADMINISTRATION||CITY OF TACOM...	5.304197e+10	0.90
166434	5YJSA1E27H	Pierce	Gig Harbor	WA	98332.0	2017	TESLA	MODEL S	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	210	0	26	169045789	POINT (-122.589645 47.342345)	BONNEVILLE POWER ADMINISTRATION||CITY OF TACOM...	5.305307e+10	2.00

166435 rows × 18 columns

"""Cleaning Legislative District dataset"""
# Removing the 0 in front of the single digit legislative districts, then making a copy of the data frame with only certain columns
leg['Legislative District'] = leg['Legislative District'].str.lstrip('0')
subset = leg[['Legislative District', 'Total Population 2010', 'White Alone 2010', 'Black or African American Alone 2010', 
             'American Indian and Alaska Native Alone 2010', 'Asian Alone 2010', 'Native Hawaiian and Other Pacific Islander Alone 2010',
             'Some Other Race Alone 2010', 'Population of Two or More Races 2010']].copy()
subset

	Legislative District	Total Population 2010	White Alone 2010	Black or African American Alone 2010	American Indian and Alaska Native Alone 2010	Asian Alone 2010	Native Hawaiian and Other Pacific Islander Alone 2010	Some Other Race Alone 2010	Population of Two or More Races 2010
0	Washington State	6724540	5196362	240042	103869	481067	40475	349799	312926
1	10	137233	119541	2004	1948	4275	513	3870	5082
2	11	137278	71456	15773	1122	30275	1384	9082	8186
3	12	137285	107156	467	4599	990	175	20029	3869
4	13	137280	108807	1344	1434	1640	131	19459	4465
5	14	137218	101108	1410	6702	1792	133	20918	5155
6	15	137224	81973	1047	4541	1011	106	43623	4923
7	16	137198	99538	1966	1241	1484	257	28740	3972
8	17	137230	114836	3088	1160	6967	1195	3936	6048
9	18	137217	122591	1255	947	5725	354	1769	4576
10	19	137232	119320	1196	3164	2303	320	5632	5297
11	20	137204	123839	772	1929	1202	239	4704	4519
12	21	137260	97656	5213	1159	19384	782	6187	6879
13	22	137239	111921	4016	1494	8589	1115	2601	7503
14	23	137217	111882	3350	2521	8193	1106	2193	7972
15	24	137280	120267	1058	6218	1920	221	2481	5115
16	25	137235	106168	6047	2074	8050	1646	4765	8485
17	26	137274	118577	3038	1568	4438	1093	1733	6827
18	27	137224	93814	13627	2726	10136	1254	5407	10260
19	28	137216	98047	11895	1416	9357	1924	3598	10979
20	29	137257	78163	17608	2035	12565	4404	9652	12830
21	30	137261	86079	10944	1494	16655	3064	10428	8597
22	31	137214	117744	2072	3064	4177	715	3812	5630
23	32	137225	96732	7011	1236	19614	581	5090	6961
24	33	137237	76183	15672	1769	18059	3563	13675	8316
25	34	137208	96382	8523	1557	14499	1214	7648	7385
26	35	137219	119040	2032	3147	3220	638	3163	5979
27	36	137247	114573	3725	1007	9529	298	2052	6063
28	37	137192	54545	31408	1174	36167	903	4957	8038
29	38	137242	105267	4140	4619	8138	839	7044	7195
30	39	137201	120439	1414	1982	2704	353	5493	4816
31	40	137259	117134	1303	1533	3807	255	8129	5098
32	41	137250	95315	2559	402	30936	269	2270	5499
33	42	137231	115238	1195	5107	4798	367	5573	4953
34	43	137252	107212	4961	860	15187	290	2058	6684
35	44	137246	112605	2308	1070	11220	456	3301	6286
36	45	137281	108934	1660	527	17449	206	2970	5535
37	46	137253	104001	5446	934	16449	414	2899	7110
38	47	137210	90810	9587	1103	19594	1637	6625	7854
39	48	137226	91704	2784	543	32355	261	4229	5350
40	49	137244	113137	3995	1370	4707	1142	6630	6263
41	1	137236	108239	2426	865	15785	417	3564	5940
42	2	137202	110085	4987	2254	5768	1532	3762	8814
43	3	137226	116887	3756	3294	3436	810	2036	7007
44	4	137281	126031	1273	1485	2366	298	1636	4192
45	5	137210	116500	1567	865	10306	275	2330	5367
46	6	137198	121721	2681	1840	3479	694	1784	4999
47	7	137263	121820	647	6208	1143	199	2746	4500
48	8	137202	113603	2047	1200	4323	223	10628	5178
49	9	137223	111742	1745	1362	4901	210	12888	4375

"""Legislative District racial distributions"""
# Selects the race columns in the subset dataframe and creates a new column for each race that calculates the percent of that race out
# of the total population in each legislative district. Rounds the percent to two decimal points
for column in subset.columns[2:9]:
    subset.loc[:, column + ' Percentage'] = (subset[column] / subset['Total Population 2010']) * 100
    subset = subset.round(2)
    
# Testing the race percentage calculations on a small subset of the dataframe
subset_test = subset[subset['Legislative District'].isin(['10', '11', '12', '13', '14'])]
columns = ['Legislative District', 'White Alone 2010 Percentage', 'Black or African American Alone 2010 Percentage', 
           'American Indian and Alaska Native Alone 2010 Percentage', 'Asian Alone 2010 Percentage', 
           'Native Hawaiian and Other Pacific Islander Alone 2010 Percentage', 'Some Other Race Alone 2010 Percentage', 
           'Population of Two or More Races 2010 Percentage']
subset_test = subset_test.reset_index(drop=True)
expected_output2 = pd.DataFrame({'Legislative District': ['10', '11', '12', '13', '14'], 'White Alone 2010 Percentage': [87.11, 52.05, 78.05, 79.26, 73.68], 
                                 'Black or African American Alone 2010 Percentage': [1.46, 11.49, 0.34, 0.98, 1.03], 
                                 'American Indian and Alaska Native Alone 2010 Percentage': [1.42, 0.82, 3.35, 1.04, 4.88], 
                                 'Asian Alone 2010 Percentage': [3.12, 22.05, 0.72, 1.19, 1.31], 
                                 'Native Hawaiian and Other Pacific Islander Alone 2010 Percentage': [0.37, 1.01, 0.13, 0.10, 0.10], 
                                 'Some Other Race Alone 2010 Percentage': [2.82, 6.62, 14.59, 14.17, 15.24], 
                                 'Population of Two or More Races 2010 Percentage': [3.70, 5.96, 2.82, 3.25, 3.76]})
assert subset_test[columns].equals(expected_output2), 'Test failed'
print('Test passed')
subset

Test passed

	Legislative District	Total Population 2010	White Alone 2010	Black or African American Alone 2010	American Indian and Alaska Native Alone 2010	Asian Alone 2010	Native Hawaiian and Other Pacific Islander Alone 2010	Some Other Race Alone 2010	Population of Two or More Races 2010	White Alone 2010 Percentage	Black or African American Alone 2010 Percentage	American Indian and Alaska Native Alone 2010 Percentage	Asian Alone 2010 Percentage	Native Hawaiian and Other Pacific Islander Alone 2010 Percentage	Some Other Race Alone 2010 Percentage	Population of Two or More Races 2010 Percentage
0	Washington State	6724540	5196362	240042	103869	481067	40475	349799	312926	77.27	3.57	1.54	7.15	0.60	5.20	4.65
1	10	137233	119541	2004	1948	4275	513	3870	5082	87.11	1.46	1.42	3.12	0.37	2.82	3.70
2	11	137278	71456	15773	1122	30275	1384	9082	8186	52.05	11.49	0.82	22.05	1.01	6.62	5.96
3	12	137285	107156	467	4599	990	175	20029	3869	78.05	0.34	3.35	0.72	0.13	14.59	2.82
4	13	137280	108807	1344	1434	1640	131	19459	4465	79.26	0.98	1.04	1.19	0.10	14.17	3.25
5	14	137218	101108	1410	6702	1792	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
6	15	137224	81973	1047	4541	1011	106	43623	4923	59.74	0.76	3.31	0.74	0.08	31.79	3.59
7	16	137198	99538	1966	1241	1484	257	28740	3972	72.55	1.43	0.90	1.08	0.19	20.95	2.90
8	17	137230	114836	3088	1160	6967	1195	3936	6048	83.68	2.25	0.85	5.08	0.87	2.87	4.41
9	18	137217	122591	1255	947	5725	354	1769	4576	89.34	0.91	0.69	4.17	0.26	1.29	3.33
10	19	137232	119320	1196	3164	2303	320	5632	5297	86.95	0.87	2.31	1.68	0.23	4.10	3.86
11	20	137204	123839	772	1929	1202	239	4704	4519	90.26	0.56	1.41	0.88	0.17	3.43	3.29
12	21	137260	97656	5213	1159	19384	782	6187	6879	71.15	3.80	0.84	14.12	0.57	4.51	5.01
13	22	137239	111921	4016	1494	8589	1115	2601	7503	81.55	2.93	1.09	6.26	0.81	1.90	5.47
14	23	137217	111882	3350	2521	8193	1106	2193	7972	81.54	2.44	1.84	5.97	0.81	1.60	5.81
15	24	137280	120267	1058	6218	1920	221	2481	5115	87.61	0.77	4.53	1.40	0.16	1.81	3.73
16	25	137235	106168	6047	2074	8050	1646	4765	8485	77.36	4.41	1.51	5.87	1.20	3.47	6.18
17	26	137274	118577	3038	1568	4438	1093	1733	6827	86.38	2.21	1.14	3.23	0.80	1.26	4.97
18	27	137224	93814	13627	2726	10136	1254	5407	10260	68.37	9.93	1.99	7.39	0.91	3.94	7.48
19	28	137216	98047	11895	1416	9357	1924	3598	10979	71.45	8.67	1.03	6.82	1.40	2.62	8.00
20	29	137257	78163	17608	2035	12565	4404	9652	12830	56.95	12.83	1.48	9.15	3.21	7.03	9.35
21	30	137261	86079	10944	1494	16655	3064	10428	8597	62.71	7.97	1.09	12.13	2.23	7.60	6.26
22	31	137214	117744	2072	3064	4177	715	3812	5630	85.81	1.51	2.23	3.04	0.52	2.78	4.10
23	32	137225	96732	7011	1236	19614	581	5090	6961	70.49	5.11	0.90	14.29	0.42	3.71	5.07
24	33	137237	76183	15672	1769	18059	3563	13675	8316	55.51	11.42	1.29	13.16	2.60	9.96	6.06
25	34	137208	96382	8523	1557	14499	1214	7648	7385	70.25	6.21	1.13	10.57	0.88	5.57	5.38
26	35	137219	119040	2032	3147	3220	638	3163	5979	86.75	1.48	2.29	2.35	0.46	2.31	4.36
27	36	137247	114573	3725	1007	9529	298	2052	6063	83.48	2.71	0.73	6.94	0.22	1.50	4.42
28	37	137192	54545	31408	1174	36167	903	4957	8038	39.76	22.89	0.86	26.36	0.66	3.61	5.86
29	38	137242	105267	4140	4619	8138	839	7044	7195	76.70	3.02	3.37	5.93	0.61	5.13	5.24
30	39	137201	120439	1414	1982	2704	353	5493	4816	87.78	1.03	1.44	1.97	0.26	4.00	3.51
31	40	137259	117134	1303	1533	3807	255	8129	5098	85.34	0.95	1.12	2.77	0.19	5.92	3.71
32	41	137250	95315	2559	402	30936	269	2270	5499	69.45	1.86	0.29	22.54	0.20	1.65	4.01
33	42	137231	115238	1195	5107	4798	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61
34	43	137252	107212	4961	860	15187	290	2058	6684	78.11	3.61	0.63	11.07	0.21	1.50	4.87
35	44	137246	112605	2308	1070	11220	456	3301	6286	82.05	1.68	0.78	8.18	0.33	2.41	4.58
36	45	137281	108934	1660	527	17449	206	2970	5535	79.35	1.21	0.38	12.71	0.15	2.16	4.03
37	46	137253	104001	5446	934	16449	414	2899	7110	75.77	3.97	0.68	11.98	0.30	2.11	5.18
38	47	137210	90810	9587	1103	19594	1637	6625	7854	66.18	6.99	0.80	14.28	1.19	4.83	5.72
39	48	137226	91704	2784	543	32355	261	4229	5350	66.83	2.03	0.40	23.58	0.19	3.08	3.90
40	49	137244	113137	3995	1370	4707	1142	6630	6263	82.43	2.91	1.00	3.43	0.83	4.83	4.56
41	1	137236	108239	2426	865	15785	417	3564	5940	78.87	1.77	0.63	11.50	0.30	2.60	4.33
42	2	137202	110085	4987	2254	5768	1532	3762	8814	80.24	3.63	1.64	4.20	1.12	2.74	6.42
43	3	137226	116887	3756	3294	3436	810	2036	7007	85.18	2.74	2.40	2.50	0.59	1.48	5.11
44	4	137281	126031	1273	1485	2366	298	1636	4192	91.81	0.93	1.08	1.72	0.22	1.19	3.05
45	5	137210	116500	1567	865	10306	275	2330	5367	84.91	1.14	0.63	7.51	0.20	1.70	3.91
46	6	137198	121721	2681	1840	3479	694	1784	4999	88.72	1.95	1.34	2.54	0.51	1.30	3.64
47	7	137263	121820	647	6208	1143	199	2746	4500	88.75	0.47	4.52	0.83	0.14	2.00	3.28
48	8	137202	113603	2047	1200	4323	223	10628	5178	82.80	1.49	0.87	3.15	0.16	7.75	3.77
49	9	137223	111742	1745	1362	4901	210	12888	4375	81.43	1.27	0.99	3.57	0.15	9.39	3.19

"""Merging EV and Legislative District datasets"""
ev_leg = pd.merge(ev, subset, on='Legislative District')
ev_leg

	VIN (1-10)	County	City	State	Postal Code	Model Year	Make	Model	Electric Vehicle Type	Clean Alternative Fuel Vehicle (CAFV) Eligibility	...	Native Hawaiian and Other Pacific Islander Alone 2010	Some Other Race Alone 2010	Population of Two or More Races 2010	White Alone 2010 Percentage	Black or African American Alone 2010 Percentage	American Indian and Alaska Native Alone 2010 Percentage	Asian Alone 2010 Percentage	Native Hawaiian and Other Pacific Islander Alone 2010 Percentage	Some Other Race Alone 2010 Percentage	Population of Two or More Races 2010 Percentage
0	3C3CFFGE4E	Yakima	Yakima	WA	98902.0	2014	FIAT	500	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
1	5YJ3E1EA5L	Yakima	Yakima	WA	98902.0	2020	TESLA	MODEL 3	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
2	JA4T5VA98P	Yakima	Yakima	WA	98908.0	2023	MITSUBISHI	OUTLANDER	Plug-in Hybrid Electric Vehicle (PHEV)	Clean Alternative Fuel Vehicle Eligible	...	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
3	5YJ3E1EA7K	Yakima	Tieton	WA	98947.0	2019	TESLA	MODEL 3	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
4	1C4JJXR64N	Yakima	Naches	WA	98937.0	2022	JEEP	WRANGLER	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	...	133	20918	5155	73.68	1.03	4.88	1.31	0.10	15.24	3.76
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
166430	KNDC3DLC2P	Whatcom	Bellingham	WA	98225.0	2023	KIA	EV6	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	...	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61
166431	1N4AZ1CP6J	Whatcom	Ferndale	WA	98248.0	2018	NISSAN	LEAF	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61
166432	KM8K33AG8L	Whatcom	Bellingham	WA	98226.0	2020	HYUNDAI	KONA	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61
166433	7SAXCAE56P	Whatcom	Blaine	WA	98230.0	2023	TESLA	MODEL X	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	...	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61
166434	1C4RJXN60R	Whatcom	Bellingham	WA	98225.0	2024	JEEP	WRANGLER	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	...	367	5573	4953	83.97	0.87	3.72	3.50	0.27	4.06	3.61

166435 rows × 33 columns

def merge_data(csv1_path, csv2_path):
    ev_data = pd.read_csv(csv1_path)
    racial_data = pd.read_csv(csv2_path)

    ev_racial_data = ev_data.merge(racial_data, how="left", left_on="Legislative District", right_on="Legislative District")
    ev_racial_data = ev_racial_data.drop(["Unnamed: 0_x", "Unnamed: 0_y"], axis=1) # noticed the merge automatically includes
    # these columns so we need to remove them

    return ev_racial_data

ev_racial_data = merge_data("Clean_EV_Population.csv", "Clean_Legislative.csv")
display(ev_racial_data)
assert type(ev_racial_data) == pd.DataFrame
assert list(ev_racial_data.columns) == ['VIN (1-10)', 'County', 'City', 'State', 'Postal Code', 'Model Year',
       'Make', 'Model', 'Electric Vehicle Type',
       'Clean Alternative Fuel Vehicle (CAFV) Eligibility', 'Electric Range',
       'Base MSRP', 'Legislative District', 'DOL Vehicle ID',
       'Vehicle Location', 'Electric Utility', '2020 Census Tract', 'geometry',
       'Total Population 2010', 'Population of One Race 2010',
       'White Alone 2010', 'Black or African American Alone 2010',
       'American Indian and Alaska Native Alone 2010', 'Asian Alone 2010',
       'Native Hawaiian and Other Pacific Islander Alone 2010',
       'Some Other Race Alone 2010', 'Population of Two or More Races 2010',
       'Hispanic or Latino (of Any Race) 2010', 'Not Hispanic or Latino 2010',
       'Population of One Race Not Hispanic or Latino 2010',
       'White Alone Not Hispanic or Latino 2010',
       'Black or African American Alone Not Hispanic or Latino 2010',
       'American Indian and Alaska Native Alone Not Hispanic or Latino 2010',
       'Asian Alone Not Hispanic or Latino 2010',
       'Native Hawaiian and Other Pacific Islander Alone Not Hispanic or Latino 2010',
       'Some Other Race Alone Not Hispanic or Latino 2010',
       'Population of Two or More Races Not Hispanic or Latino 2010',
       'Total Population Over 18 2010', 'Population of One Race Over 18 2010',
       'White Alone Over 18 2010',
       'Black or African American Alone Over 18 2010',
       'American Indian and Alaska Native Alone Over 18 2010',
       'Asian Alone Over 18 2010',
       'Native Hawaiian and Other Pacific Islander Alone Over 18 2010',
       'Some Other Race Alone Over 18 2010',
       'Population of Two or More Races Over 18 2010',
       'Hispanic or Latino (of Any Race) Over 18 2010',
       'Not Hispanic or Latino Over 18 2010',
       'Population of One Race Not Hispanic or Latino Over 18 2010',
       'White Alone Not Hispanic or Latino Over 18 2010',
       'Black or African American Alone Not Hispanic or Latino Over 18 2010',
       'American Indian and Alaska Native Alone Not Hispanic or Latino Over 18 2010',
       'Asian Alone Not Hispanic or Latino Over 18 2010',
       'Native Hawaiian and Other Pacific Islander Alone Not Hispanic or Latino Over 18 2010',
       'Some Other Race Alone Not Hispanic or Latino Over 18 2010',
       'Population of Two or More Races Not Hispanic or Latino Over 18 2010'
]
assert len(ev_racial_data) == 166790

	VIN (1-10)	County	City	State	Postal Code	Model Year	Make	Model	Electric Vehicle Type	Clean Alternative Fuel Vehicle (CAFV) Eligibility	...	Hispanic or Latino (of Any Race) Over 18 2010	Not Hispanic or Latino Over 18 2010	Population of One Race Not Hispanic or Latino Over 18 2010	White Alone Not Hispanic or Latino Over 18 2010	Black or African American Alone Not Hispanic or Latino Over 18 2010	American Indian and Alaska Native Alone Not Hispanic or Latino Over 18 2010	Asian Alone Not Hispanic or Latino Over 18 2010	Native Hawaiian and Other Pacific Islander Alone Not Hispanic or Latino Over 18 2010	Some Other Race Alone Not Hispanic or Latino Over 18 2010	Population of Two or More Races Not Hispanic or Latino Over 18 2010
0	3C3CFFGE4E	Yakima	Yakima	WA	98902.0	2014	FIAT	500	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	21616	79487	77744	71356	907	3979	1312	85	105	1743
1	5YJXCBE40H	Thurston	Olympia	WA	98513.0	2017	TESLA	MODEL X	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	6102	93558	90163	79905	3364	1366	4367	1002	159	3395
2	3MW39FS03P	King	Renton	WA	98058.0	2023	BMW	330E	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	...	11908	92174	88859	53220	10828	621	23089	944	157	3315
3	7PDSGABA8P	Snohomish	Bothell	WA	98012.0	2023	RIVIAN	R1S	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	...	8755	96811	94026	74239	3605	738	14743	548	153	2785
4	5YJ3E1EB8L	King	Kent	WA	98031.0	2020	TESLA	MODEL 3	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	15898	89387	86183	58131	10853	1010	13708	2256	225	3204
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
166785	3FA6P0SU4D	Spokane	Spokane	WA	99223.0	2013	FORD	FUSION	Plug-in Hybrid Electric Vehicle (PHEV)	Not eligible due to low battery range	...	4280	102421	100196	93650	2031	1227	2748	431	109	2225
166786	5YJYGDEE5M	King	Sammamish	WA	98074.0	2021	TESLA	MODEL Y	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	...	5316	96646	94492	79983	1214	331	12492	144	328	2154
166787	7SAYGDEE5N	Snohomish	Mukilteo	WA	98275.0	2022	TESLA	MODEL Y	Battery Electric Vehicle (BEV)	Eligibility unknown as battery range has not b...	...	8755	96811	94026	74239	3605	738	14743	548	153	2785
166788	1G1RH6E43D	Lewis	Mossyrock	WA	98564.0	2013	CHEVROLET	VOLT	Plug-in Hybrid Electric Vehicle (PHEV)	Clean Alternative Fuel Vehicle Eligible	...	5858	98475	96452	93519	467	1305	918	166	77	2023
166789	5YJSA1E27H	Pierce	Gig Harbor	WA	98332.0	2017	TESLA	MODEL S	Battery Electric Vehicle (BEV)	Clean Alternative Fuel Vehicle Eligible	...	4927	101989	98964	91180	2367	1046	3480	758	133	3025

166790 rows × 56 columns

Results

In terms of the first and second research questions, we can see that accessibility to charging stations in rural areas, especially Eastern WA, is an issue. In the urban areas, otherwise known as very populated regions, there is a surplus of charging stations. However, when you look at other regions, like the lower left corner of the state or the middle or eastern side of the state, you can see a lack of charging infrastructure. From the interactive plotly map of Washington State shown below, the blue dots represent the electric vehicles in Washington state, while the red dots represent the charging station locations. One surprising takeaway from the map is that accessibility does not seem to be a deterrent in purchasing and owning electric vehicles because if it was, there would not be blue dots where there are no red dots. Regarding the fourth research question, you can see that there are clusters of electric vehicle adoption in urban areas especially, like the greater Seattle area cities, but there is also adoption of electric vehicles all over the state. Relative to charging stations, this is not a big deal as described above, which means there are most likely many electric vehicle owners with access to private charging stations.

geometry1 = ev_racial_data["Vehicle Location"].apply(wkt.loads)
ev_racial_gdf = gpd.GeoDataFrame(ev_racial_data, geometry=geometry1, crs="EPSG:4326")

geometry2 = [Point(xy) for xy in zip(charging_df["Longitude"], charging_df["Latitude"])]
charging_gdf = gpd.GeoDataFrame(charging_df, geometry=geometry2, crs="EPSG:4326")

# Set CRS for GeoDataFrames to EPSG:4326
state_data.crs = 'EPSG:4326'

entire_state = state_data[["geometry"]].dissolve()
fig, ax = plt.subplots(figsize=(10, 5))
entire_state.plot(ax=ax, color="#EEE")
state_data.plot(ax=ax, facecolor="#CCCCCC" , edgecolor="black")

# Plot EV ownership
wa_data = gpd.sjoin(ev_racial_gdf, state_data, how='inner', predicate='intersects')
wa_data.plot(ax=ax, color="blue", marker="o", markersize=1, label="EV Ownership")

# Plot EV charging stations
charging_gdf.plot(ax=ax, color="red", marker="o", markersize=1, label="EV Charging Stations")

# Set title and turn off axis
ax.set_title("Washington with EV Charging Stations & EV Ownership")
ax.set_axis_off()

ax.legend()

plt.show()

assert type(ev_racial_gdf) == gpd.GeoDataFrame
assert type(charging_gdf) == gpd.GeoDataFrame
assert ax.get_title() == "Washington with EV Charging Stations & EV Ownership", "title does not match expected"
assert not ax.axison, "borders and labels must be hidden"
assert ax.legend() is not None, "missing legend"

# Perform spatial join between ev_racial_gdf and state_data
wa_data = gpd.sjoin(ev_racial_gdf, state_data, how='inner', predicate='intersects')

# Plot EV charging stations and EV ownership using Plotly Express
fig = px.scatter_mapbox(charging_gdf,
                        lat="Latitude",
                        lon="Longitude",
                        color_discrete_sequence=["red"],
                        hover_name="Station Name",
                        mapbox_style="carto-positron",
                        zoom=5,
                        opacity=0.7,
                        title="Washington with EV Charging Stations & EV Ownership")

fig.add_scattermapbox(lat=wa_data.geometry.y,
                      lon=wa_data.geometry.x,
                      mode="markers",
                      marker=dict(color="blue", size=4),
                      hoverinfo="text",
                      text="EV Home",
                      name="EV Ownership")

# Using renderer to display interactive map in Jupyter Notebook
fig.show(renderer='iframe')

# Show the interactive map
fig.show()

One of the research questions that we wanted to address was “Is there a racial disparity in access to electric vehicles in Washington?” We first investigated this by creating a scatter plot that compares the percentage of white individuals within a legislative district to the percentage of EVs in a legislative district out of all the EVs registered in Washington. Overall, there was not a significant relationship between the two variables, but there did appear to be a slight positive correlation. For example, Legislative District 15, which includes portions of Adams, Benton, Franklin, Grant, and Yakima counties in southern Washington, is 60% white and makes up only 0.2% of all EVs in the state, while Legislative District 5, which includes portions of Kittias, Maple Valley, Renton, Issaquah, and Snohomish counties in central Washington, is 85% white and makes up 4% of all EVs in the state. This is a significantly higher percentage of EV ownership with a higher proportion of white individuals. From these results, we can conclude that race may be a limiting factor in access to electric vehicle ownership. Yakima County, for example, located in Legislative District 15, is Washington State’s largest majority-Hispanic county, as of 2020 (Wikipedia). Due to systematic housing discrimination, people of color experience lower rates of homeownership and lower average income, which limits access to electric vehicles. This is due to the cost of buying and maintaining an electric vehicle, which already has a higher upfront cost than gas-powered vehicles, and due to the limited public charging infrastructure available, especially in less-developed areas. However, it is important to note that race is not the only factor that impacts access to EVs. As shown in the scatterplot, many legislative districts displayed are over 85% white and have similar EV ownership rates to legislative districts with a lower white population. This highlights that other factors come into play when addressing the lack of electric vehicle adoption today, including the lack of robust charging infrastructure and consumers’ views on affordability and EV performance.

"""Interactive scatterplot comparing the percentage of white individuals (x) to the percentage of EV ownership in each
legislative district in WA"""
# Uses the plotlyexpress library to create the scatterplot
interactive = px.scatter(ev_leg, x='White Alone 2010 Percentage', y='Percentage EV Ownership', hover_data=['Legislative District'])
interactive.update_layout(title='Percentage of White Individuals vs Percentage of EV Ownership per WA Legislative District', 
                          xaxis_title='Percentage of White Individuals', 
                          yaxis_title='Percentage of EV Ownership')
# Using renderer to display interactive graph in Jupyter Notebook
interactive.show(renderer='iframe')
interactive.show()

To further explore the question of whether there is a racial disparity in access to electric vehicles in Washington, we also created two bar plots that compared the racial distribution of the legislative districts with the highest and lowest EV ownership in Washington. The results are not drastically different, as in both legislative districts, the population is majority white. However, an interesting result from the bar plot showing Legislative District 15, which has the lowest percentage of EV ownership, was that over 30% of the population identified themselves as “Some Other Race.” The United States Census Bureau analyzed what the most common groups were that identified themselves as this in the racial category, and they concluded that the top 3 most common groups were Brazilians, Guyanese, and Cabo Verdeans. Additionally, a large majority of census respondents who selected “Some Other Race” identified as Hispanic when asked to disclose their ethnicity. This aligns with the fact that Yakima County, which as mentioned before is within Legislative District 15, is the largest majority-Hispanic county in Washington. It is meaningful to note that the legislative district with the lowest EV ownership in the state has a sizable proportion of people of color, specifically Hispanic or Latino individuals, which further supports the idea that communities of color may face a greater lack of access to electric vehicles than a community that is predominantly white, such as one found in Legislative District 41. Legislative District 41 includes Mercer Island, Newcastle, Bellevue, Renton, Issaquah, and Sammamish. Bellevue, for example, is a tech company hub and includes suburbs of higher socioeconomic status and predominantly white communities. Electric vehicles appear to be more available in places like Bellevue, where affordability and accessibility may be more feasible. Overall, while there wasn’t a strong relationship between race and EV ownership shown by the visualizations we created, it is important to note how race can be a factor in accessibility of electric vehicles.

"""Barplots showing racial distribution of legislative districts with highest and lowest percent EV ownership"""
# Locates the row with the highest and lowest percent EV ownership
high_ev_index = ev_leg['Percentage EV Ownership'].idxmax()
high_ev_district = ev_leg.loc[high_ev_index]
low_ev_index = ev_leg['Percentage EV Ownership'].idxmin()
low_ev_district = ev_leg.loc[low_ev_index]

# Creating dictionaries with the columns used to make the barplots
high_ev_data = {'Legislative District': high_ev_district['Legislative District'], 'Race': ['White', 'Black', 'Asian', 'Two or More Races', 'Some Other Race'],
                'Percentage': [high_ev_district['White Alone 2010 Percentage'], high_ev_district['Black or African American Alone 2010 Percentage'], 
                               high_ev_district['Asian Alone 2010 Percentage'], high_ev_district['Population of Two or More Races 2010 Percentage'],
                               high_ev_district['Some Other Race Alone 2010 Percentage']]}
low_ev_data = {'Legislative District': low_ev_district['Legislative District'], 'Race': ['White', 'Black', 'Asian', 'Two or More Races', 'Some Other Race'],
               'Percentage': [low_ev_district['White Alone 2010 Percentage'], low_ev_district['Black or African American Alone 2010 Percentage'], 
                              low_ev_district['Asian Alone 2010 Percentage'], low_ev_district['Population of Two or More Races 2010 Percentage'], 
                              low_ev_district['Some Other Race Alone 2010 Percentage']]}

# Converting the dictionaries into dataframes
high_ev_df = pd.DataFrame(high_ev_data)
low_ev_df = pd.DataFrame(low_ev_data)

# Creating the highest EV ownership race distribution barplot
sns.set(style='white')
sns.set_palette('Set2')
high_barplot = sns.catplot(data=high_ev_df, x='Legislative District', y='Percentage', hue='Race', kind='bar')
plt.xlabel('Legislative District 41')
plt.ylabel('Percentage of Race')
plt.title('Race Dist. of Legislative District with the Highest EV Ownership in WA')
# Making sure that the y-axes on both graphs are equal
plt.ylim(0, max(high_ev_df['Percentage']))
plt.show()

# Creating the lowest EV ownership race distribution barplot
sns.set(style='white')
sns.set_palette('Set2')
low_barplot = sns.catplot(data=low_ev_df, x='Legislative District', y='Percentage', hue='Race', kind='bar')
plt.xlabel('Legislative District 15')
plt.ylabel('Percentage of Race')
plt.title('Race Dist. of Legislative District with the Lowest EV Ownership in WA')
# Making sure that the y-axes on both graphs are equal
plt.ylim(0, max(high_ev_df['Percentage']))
plt.show()

Impact and Limitations

On a more general level, from this report we may be assuming that everyone understands the definition of an electric vehicle and what qualifies to be an electric vehicle. Although this may seem ridiculous, electric vehicles are newer and still developing which makes it difficult to form a solid way to categorize and define them. There are many different ways to apply these findings such as policy creation, environmental impacts, and economical decisions. For example, these findings could be useful for policymakers about where Washington state is at in terms of electric vehicle adoption and availability or accessibility of charging stations for the owners. This could influence investments made by the city or the local government on whether or not they should build more charging infrastructure based on the demand. In terms of environment, understanding the distribution of electric vehicles across different regions of Washington state can shed light on the potential greenhouse gas emissions reduction and air pollution. To transition, this could be helpful for anyone who may seem to raise an eyebrow at the importance of sustainability, like big companies who are actually trying to make a positive impact on the environment and are not just greenwashing. Any companies involved in the electric vehicle industry could benefit with insights into market demand for electric vehicles or chargers as well as identifying opportunities for growth and further investment.

Climate change and the effects of it are a collective issue. However, it is important to highlight how certain groups, based on discriminative history, feel these impacts much greater than others. By first diving into the idea of a racial disparity in access to electric vehicles, we are able to communicate an issue of environmental injustice. Communities deserve equitable access to sustainable practices, such as purchasing an electric vehicle. Many blame those who still use gas-fueled vehicles as careless towards the environment, but it is important to highlight the inaccessibility of electric vehicles for some groups. Research like this can serve as a basis for building federal policy and incentive which achieves equitable access for all. As with any topic regarding environmental justice, it is important to get involved with these communities to hear their own views on electric vehicle access. While data can begin to tell the story of injustice, the story is not fully told until the ones experiencing the injustice are heard and uplifted. It may harm these communities to, for example, release this data claiming there is a racial disparity, all while ignoring how people of color really feel about these issues. It’s important to highlight that there were significant limitations with our data setting. For example, the 2010 Census Data by Legislative District is 14 years old, and population distributions have likely changed significantly. Trying to compare this data to the electric vehicle dataset, which is present day data, can be inaccurate and lead to conclusions that do not represent the state of electric vehicle adoption and racial distribution today. Additionally, the visualizations created to dive into the topic of racial disparity in access to electric vehicles did not have very obvious conclusions. While we claim here that a disparity may exist, it’s important to understand that the analysis does not claim a tight relationship between the two variables. Even so, we still believe it is important to highlight how a sustainable practice hailed by many, such as electric vehicles, can uncover issues of equity and discrimination. Other maybe more minor limitations from our data include the accuracy and completeness of the data, since some of the data may not be completely up to date regarding both the electric vehicle populations and newly developed charging stations. In order to use this report to make informed decisions, we may need to find a very recently updated dataset with electric vehicles and charging stations that is updated on a monthly basis.

Challenge Goals

Multiple datasets: In our code, we use three different datasets: current registered electric vehicles in Washington, racial demographics in 2010 by Legislative District, and current electric vehicle charging stations in Washington state. We merged the electric vehicles and Legislative District datasets on the Legislative District column and used the geometries of the electric vehicle locations and charging station datasets to create maps of Washington state. This challenge goal was a bit difficult to achieve, especially the part where all three datasets needed to be used together in some way. These datasets were not perfectly cohesive with each other and therefore made them difficult to work with together, but Sophia was able to create a map that incorporates all three datasets to convey EV and charging station distribution, and then being able to identify the legislative district by color.

New library: Our project meets this goal because we used Plotly to make an interactive scatterplot (i.e., being able to hover the cursor over a point and see location). We focused on learning an interactive visualization library on a scatterplot rather than a map like we originally planned because it seemed more feasible in our time frame and the learning curve of using Plotly Express was pretty straightforward and felt similar to graphing with seaborn or matplotlib.

Plan Evaluation

In the team project proposal, our written plan was not too detailed, but after meeting with our assigned mentor, he suggested that we sit down and make a rough timeline of when we want to complete each of the steps we had listed. Our expectations for completing the data organizing and data cleaning was only a day or two, which was what we found to be very doable. Looking back, I would say we did accomplish some other data cleaning tasks throughout the completion of the code, as the more we worked with the datasets, the easier it was to discover certain changes and cleaning steps necessary. The next step listed, merging the datasets, took much longer than we thought it would. When merging the legislation dataset containing all of the racial demographic information and the electric vehicle population dataset, it was difficult to merge them on the Legislative District column. The electric vehicle population dataset’s column named “Legislative District” was a float type, while the racial demographic dataset’s issue was that the Legislative Districts 1 through 9 (single digits) had a leading 0 in front of the number to make them 2 digits instead of one which needed to be removed. Additionally, for the maps created of Washington state, we took the state dataset given to us in the Mapping Assessment to map the background of Washington state along with its census tracts to help get a better understanding and perspective on the areas in the state relative to where the electric vehicle ownership is and relative to where the EV charging stations are located. However, this portion of the project and code was one of the lengthiest and most time consuming. We made attempts at a folium map, but it only “ran” once before crashing the Jupyter notebook and Google Colab so we decided to scratch it and use solely the plotly. Constantly, we were frustrated with errors and discovered that there were issues with the geometry columns in the datasets not actually representing geometries and instead were object types or floats. Therefore, we had to change the types of the columns as well as make the charging station dataset and merged dataset, containing the information on electric vehicles and racial demographics, into a geodataframe. After resolving the issues of the geometry columns, we were able to map everything. Later on, we decided to perform some calculations on percentages (such as the racial demographics and the percentage of EV ownership) as a lot of the data was raw population numbers or difficult to interpret from the original dataset, so calculating percentages seemed to be a lot more helpful for interpreting results and making visualizations. This required doing simple percentage calculations and then creating columns to then merge onto the main dataset we were working on for data analysis.

Testing

Calculations were performed to see the percentage of EV ownership and the racial distributions in the census data. In order to test that these calculations were correct, assert statements were used. We created an ‘expected output’ dataframe based on a small subset of the rates dataframe (which calculated the percentage of EV ownership) and the subset data frame (which calculated the racial distributions), and then hand-calculated the percentages. We then used the pandas function ‘.equals’ to assert that a subset of the two data frames mentioned matched our hand-calculations. For the merging of the datasets function, we were able to set up test assertions for inclusion of all correct columns by using the df.columns for each individual dataset. Then, the length assertion would be for the number of rows and just a simple initial assert ensuring that the dataframe imported was a pandas dataframe. For the mapping, we decided to use some of the testing that was given to us in the Mapping Assessment. Here, we tested for correct title names, presence of a legend, missing borders and labels. Also, we tested the GeoDataFrames that were created in spite of the mapping to ensure that they were GeoDataFrames.

Collaboration

Besides our peers and helpful teaching staff, the majority of our mapping section was completed thanks to the already completed mapping assessment we turned in a few weeks ago. Using some of the sources for importing the state data and the coding chunks used to sort the data just for Washington state, we were able to include this basic background or outline of the state and its census tracts in our own creation of maps for the electric vehicles and charging stations. We used generative AI in some portions of the mapping for miniscule tasks like changing map fill color and changing the marker size in the plot. Generative AI was also used to resolve some crazy unknown key errors or type errors that a simple Google search could not resolve for us. Some other resources that were available to our group was the help of using plotly in R Studio for one of our group members' informatics class. Although the coding lines look different in Python than R Studio, it was easier to understand and develop the plotly interactive map having used it all quarter in another class. Generative AI was also helpful for little issues we were running into with the merging of the census and EV data. For example, we kept running into an error when calculating the percentage of EV ownership where duplicate columns would be made which caused a merge error down the line when edits were made. We prompted Generative AI with the error which suggested using an if statement to drop the duplicate column if it is detected in the dataset, which solved the problem. Another way Generative AI was helpful was in fixing the barplot visualization for the racial distribution. We wanted the y-axes to have the same range to avoid a misleading visualization, and Generative AI suggested using the function ylim, which we read the documentation for in order to understand how it worked. Additionally, referring to the documentation for libraries like pandas, seaborn, and plotly was incredibly helpful in figuring out what certain functions would be helpful when merging datasets.